Oracle Teach Skill

Transform mature knowledge into teachable materials

Purpose

Oracle-teach generates learning materials from patterns that have reached 🌿 Pattern level or higher. It's the "สร้างคน" output — taking what we've learned and making it transferable.

Proactive Triggers

MUST Use Teach When:

Knowledge Transfer:

• User says: "teach me about", "explain the pattern"
• User says: "how does X work", "why do we do X"
• User says: "create a guide for", "document this"

Team Sharing:

• Onboarding new team member
• Writing documentation
• Creating training materials

Pattern Explanation:

• User asks about established pattern
• Need to justify a decision
• Explaining "why we do it this way"

Teaching Formats

Quick Explanation

## Pattern: [Name]

**What**: One-line description
**Why**: The reasoning behind it
**How**: Steps to apply
**Example**: Concrete use case

Full Guide

# Guide: [Topic]

## Overview
What this pattern/principle is about.

## Background
Why this emerged, what problem it solves.

## The Pattern
Step-by-step application.

## Examples
### Example 1: [Context]
[Walkthrough]

### Example 2: [Context]
[Walkthrough]

## Common Mistakes
- Mistake 1: Why it's wrong
- Mistake 2: Why it's wrong

## Related Patterns
- [Link to related]

## Source
From Oracle knowledge base, maturity level: 🌿/🌳/🔮

Micro-lesson (60 seconds)

## 60-Second Lesson: [Topic]

**The Rule**: [One sentence]

**Why It Matters**: [One sentence]

**Try This**: [One action]

**Remember**: [Mnemonic or key phrase]

Teaching Workflow

1. Find Teachable Knowledge

oracle_search({
  type: "pattern",  // or principle, wisdom
  limit: 10
})
// Filter for teachable: true or maturity >= 🌿

2. Choose Format

3. Generate Material

User: "teach me about subagent delegation"

AI checks Oracle → finds 🌿 Pattern "Subagent Delegation"

AI generates:
## Pattern: Subagent Delegation

**What**: Use Haiku subagents for bulk operations, Opus for review
**Why**: Saves context tokens, parallel execution, cost efficiency
**How**:
1. Identify bulk task (5+ files, heavy search)
2. Spawn subagent with clear instructions
3. Let subagent return summary
4. Main agent reviews and acts

**Example**: Searching 50 files for a pattern
- ❌ Main agent reads all 50 files (expensive)
- ✅ Subagent searches, returns top 5 matches

Integration with Oracle Ecosystem

Teachability Criteria

Output Locations

Quick Reference